coldStart trap after "kill -HUP" of snmpd?

[yjdabear]

While working with a third-party network device that uses the UCD-SNMP-MIB implementation, I'm observing the behavior that one each of "nsNotifyShutdown" and "coldStart" trap is generated, everytime snmpd receives "kill -HUP" to pick up config modifications. Does anyone know if this is the "correct / expected / normal" behavior? I mean I could see the rationale behind it (the assumption that snmpd is down due to device reboot), but OTOH it's sort of "crying wolf" when the physical device itself didn't actually reboot. I'd like to avoid creating this kind of "false" alarm, if possible.

[Joe Clarke]

This question would be better asked on one of the Net-SNMP forums.  However, in looking at the 5.4.2.1 code, only an nsNotifyRestart trap is sent on a SIGHUP.

[yjdabear]

Ah, since NET-SNMP starts at 5.x, maybe what I see is the behavior of the UCD-SNMP 4.x or older releases.

[Joe Clarke · Accepted solution]

I don't see any trap being sent on a SIGHUP in uc-snmp 4.2.7.1.  The only coldStart trap sent by either version is on snmpd startup.

[yjdabear]

Thanks for looking into the source code! Your posts made me question my assumption it was a  "kill -HUP". It turns out snmpd was being restarted with the init script. Once I asked the vendor to issuing "kill -HUP" instead, only an nsNotifyRestart is generated.